This is incredibly sloppy work. It just won’t do for any kind of serious analysis. 

First of all, the fact that the internet has millions of nodes makes no real difference to the complexity of the system. The billions of grains of sand on a beach can be modeled as a network, but the complexity of that system isn’t particularly remarkable and the number of nodes certainly doesn’t make it comparable to the human mind. Complexity isn’t a a quantitative matter of how many things are hooked together, it is a dynamical matter of what the resulting network does.  Brains are interesting because they do interesting things, not just because they are made of lots of parts. 

This isn’t a trivial point; in fact, its the key to understanding the emergence of complex systems. A complex system is one where the behavior of the parts must be understood in the context of the system as a whole. For example, your body is a complex system. If I want to understand what your heart does, I need to understand its role in the functioning of your body as a whole. It doesn’t matter how many organs you have, what matters is what those organs contribute to your overall dynamics. The complexity of the system requires an explanation that treats your whole body as basic, and that means any investigation into the parts requires a simultaneous investigation of the whole.

The internet is also a complex system. The particular nodes in the network  are constantly changing both at the level of internet hardware and in terms of content on the web (and for what its worth, these are two nearly independent networks!) Nevertheless, the overall dynamics of the network are stable enough over time that we can identify the internet as a coherent whole. And of course we do exactly that all the time, and we’ve even given it a name: Internet. Understanding Internet as a dynamical whole is sometimes necessary for understanding the behavior and functioning of the parts, be it a packet streaming across a wire, or the rise and fall of some internet meme. 

The meme case is particularly striking. We accept and have largely grown accustomed to the way memes spread across the internet, and we’ve all seen it happen dozens of times. We tend to attribute this to the way internet works. The Kony 2012 video, for instance, was made by a particular group of people, but it spread virally because of  the overall dynamics of the internet, which no individual person could have predicted in advance and (therefore) aren’t responsible for. The spread of this meme is a function of internet as a whole. 

That’s just to say that the Internet is already an emergent complex system. That’s a completely distinct issue from whether or not the internet is conscious. The fact that Koch would confuse this issue shows just how far we have to go until even scientists develop strong intuitions about complexity. 

I don’t think consciousness exists in the sense Koch describes. Specifically, I don’t think there is any “intrinsic” properties that minds has that are inaccessible to external review and investigation. Or to put it directly, I think science is our best shot at knowledge, and until introspection is available for peer review I think we’d be wise not to trust it. Nevertheless, Koch is referring to very obvious features of human brains: they feel things, they get happy or sad, and so on. These aren’t just features of the complexity of the brain, but they are features of the very specific functions the brain carries out. In the case of being happy or sad, these functions largely have to do with regulating the balance of neurotransmitters in the brain like serotonin. Since computers (and the Internet) don’t typically have to carry out the function of chemical balances in wetware, it would be quite surprising if they seemed to exhibit similar behavior without explicit intentional design. Becoming happy or sad like a human is not the kind of thing the Internet does, and that’s not because happiness is particular mysterious, but because Internet isn’t in that line of work. 

Nevertheless, our understanding of systemic wholes is fundamentally athropocentric. Our brains are trained to understand the dynamics of human beings, and our brains our pattern seekers: we look for symmetries that we might exploit. So our brains would really really like it if we could use the same anthropic models we use for people and apply them to the dynamics of the machines. We want to know if machines can feel happy, because we know a lot of things about what happy people do, and maybe that will help us predict and control what the machine will do. 

But this is a complete misunderstanding of the kind of machine the Internet is. Internet is the self-organized system that includes all its users as nodes. So understanding the dynamics of happiness will help understand the dynamics of internet, not because the internet is happy, but because happiness influences the dynamics of people and the internet is made of people. 

In other words, if you want to know how internet feels, the correct answer to this question will look something like this:

http://twistori.com/#i_feel

That’s how the internet feels. That’s not how any one individual feels, and it doesn’t look anything like _individual_ human consciousness, because the internet is not a human being. Using this measure of how the internet feels, you can approximate the feelings of the particular individuals as they stream by; moreover, you can use this streaming data to compile an overview of the general attitude and sentiments online. This generalization to the systems level is just what it would mean for the internet to feel something. 

It is worth nothing that on my view, getting the internet to “learn to feel” is really just a matter of getting all its users to learn to feel together: that is, to come to a high-bandwidth consensus. 

Achieving consensus is precisely what it means for a human community to self-organize.